What the checker scores
Citation-readiness is not one number you can fake — it's a handful of concrete signals an answer engine uses to decide whether your page is worth reading and safe to quote. The audit grades each one and returns a breakdown, not just a grade:
- Answerable structure. Clear question-style headings with direct, self-contained answers an engine can lift verbatim.
- Structured data. JSON-LD / schema.org markup (Article, FAQPage, HowTo, Organization) that tells engines what the page is.
- Crawlability for AI. robots.txt, llms.txt, and meta directives that actually allow the AI crawlers you want citing you.
- Entity clarity. Unambiguous names, definitions, and 'what is' statements that pin down the entities on the page.
- Freshness signals. Visible and structured dates, last-updated markers, and changelog cues engines use to trust currency.
- Extractable facts. Numbers, lists, and tables in clean markup — the shapes models quote most reliably.
- Source authority. Author, citations, and outbound references that raise the page's trust for answer engines.
- Render without JS. Whether the substance is in the HTML or trapped behind JavaScript an engine may never run.
Why citation-readiness matters now
Search is shifting from ten blue links to one synthesized answer. When an AI engine answers a question, it reads a small set of sources and cites a few — and being one of those cited sources is the new front page. A beautifully ranked page that an engine can't cleanly parse, can't verify, or isn't allowed to crawl simply doesn't get quoted. Citation-readiness is how you measure whether you're in that set.
How the score is produced
When you run the audit, Ollagraph fetches the page the way an answer engine would — rendering
JavaScript when needed — then inspects its structure, schema markup, crawl directives, entities,
and extractable facts, and returns a scored breakdown with specific findings. It's the same
/v1/aeo/citation-readiness endpoint our customers run in production
and in CI; this page is just the friendly front door to it.
How to improve a weak score
- Answer the question in the first sentence under each heading — engines lift self-contained answers, not buried ones.
- Add the right schema —
FAQPage,HowTo,Article— so engines know what the page is. - Ship an
llms.txtand a cleanrobots.txtthat allow the AI crawlers you want citing you. - Put facts in markup, not images — numbers, lists, and tables in HTML are what models quote.
- Render the substance server-side so content isn't trapped behind JavaScript an engine may not run.
Want the full playbook? Read AEO vs SEO in 2026 and the AEO audits overview.
How answer engines decide — primary sources
These signals aren't folklore; they track the public standards and research that answer engines build on. The references this checker leans on:
- RFC 9309 — Robots Exclusion Protocol (IETF, 2022) — whether an AI crawler is even allowed to read your page.
- JSON-LD 1.1 (W3C Recommendation) — the structured-data format that tells engines what your page is.
- Large language models (Wikipedia) — how the models that cite you read and chunk text.
- Retrieval-Augmented Generation (Lewis et al., 2020) — why extractable structure and freshness drive which page gets quoted.